Wesley D. Snow, Co-Founder and President of Ascendix Technologies. 30+ years in software development; a pioneer in AI for Real Estate.

getty

​There’s a moment when I start to look for sales calls with commercial real estate brokers. It happens when I tell them they don’t have to open the CRM anymore, and they can just ask a question in plain English and get the answer pulled from their actual data. Something shifts. They sit up straighter. Their eyes light up. They start asking questions not out of cautious curiosity but because of genuine excitement, because they’ve been waiting for exactly this without knowing how to say it.

That reaction tells me more about where real pain is than any survey. The database has been built for years. But getting something meaningful from the pile of records requires training, admin support or more time (and patience) than a broker has between meetings. And when data lives across multiple platforms, the assembly work alone can eat an hour before the actual thinking starts.

That’s the friction AI is best positioned to remove. But not all AI is built to do it.​

​Why Data Stays Out Of Reach ​

CRM interfaces are not intuitive by design. CRMs are built to store and organize, not to answer. Running a meaningful query (which tenants have leases expiring in the next eight months, which buyers have looked at similar assets, what’s the full activity history on this account, etc.) typically means building a report, knowing which filters to apply or asking someone who does. So brokers work around it.

When the CRM answers questions instead of waiting to be navigated, brokers stop working around it.

A broker asks in plain English, by voice, on mobile, between meetings and gets the answer from their actual records. Brochures, BOVs, reports and other documents can be generated based on live data. Even such tedious things as drafting emails can now be done much faster. The broker’s task shifts from production to review. Predictive analytics and financial modeling are moving in the same direction: less rebuilding from scratch, more time spent on work that requires expertise.

The question is whether the AI actually knows your business or is guessing from the open internet.​

The AI That Doesn’t Know Your Business ​

The natural response to any data-related challenge has been to turn to ChatGPT or Claude. They’re fast, conversational, require no training program and they work well for drafting emails or summarizing documents.

The problem surfaces when a broker asks a business-specific question. Generic AI doesn’t know which tenants are active, which leases are rolling or which investors have already passed on similar assets. So the answers don’t have any grounding in reality. An answer that sounds confident but has no grounding in your actual data costs more time to unwind than the question saved.

And even when it’s connected to the firm’s actual records through a CRM software integration with Claude AI, ChatGPT or similar tools, the problem remains. Does it know what LOI and NNN mean? Does it return full information? Which part of it is hallucination? Is my data secure? So many questions with one answer—Grounded AI. ​​

Grounded AI And What It Requires ​

The firms getting real value from AI have done something specific. They have connected it to their actual records, taught it their vocabulary and set guardrails around what it can and cannot do without a human check. That takes work, and it requires someone who understands both the AI and the CRE workflow. But when it is done right, the broker gets a reliable answer, not a confident guess.

That means AI connected to live data, operating within the firm’s security protocols, respecting each user’s permissions, understanding industry terminology, LOI, NNN, FSG, lease comp, tenant requirement and minimizing hallucinations by drawing on actual records rather than general training—a lot of requirements that standard ChatGPT or Claude integration does not provide out of the box.

Grounded AI stops brokers from thinking about tools and gets them thinking about the outcomes. ​

Why Integration Is Harder Than Vendors Will Tell You ​

The integration step is harder than most vendors will admit. Not every CRM, property management system or loan origination platform was built to be connected to AI. Some are closed by design. Others offer APIs and MCPs in theory but lack the documentation or support to make integration practical without a specialist. That dependency on someone who understands both the technology and the CRE workflow is a real barrier—and it is one the industry has not fully reckoned with yet.

The firms that do close that gap are building something their competitors cannot easily copy. Not the AI itself. The AI is available to everyone. What they are building is a system that knows their business: their clients, their deals, their history, their vocabulary. That takes time, deliberate configuration and someone willing to do the integration work properly rather than bolt on a chatbot and call it done. ​

The brokers who light up on my sales calls are not reacting to a product. They are reacting to the recognition that the work they have been doing for years, every call logged, every requirement updated, every relationship tracked, could finally be working for them instead of sitting in a database waiting to be retrieved by someone who knows the right filters. That is what grounded AI makes possible. The firms that get there first will have an advantage that compounds every quarter. The ones that settle for off-the-shelf will keep getting confident answers that have nothing to do with their actual business. That gap is already opening. ​

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?